¹ Kaufmann, C. M., The Baths of Pozzuoli: A Study of the Medieval Illuminations of Peter of Eboli's Poem (Oxford 1959) 20 n. 5 and 21 n. 4, cites this work as if it were two works written by Arnaldus de Bruxella and Francesco Aretino respectively. The full title is Libellus de mirabilibus civitatis Putheolorum et locorum vicinorum ac de nominibus virtutibusque balneorum ibidem existentium. Google Scholar

² Huillard, A.-Bréholles, ‘Notice sur le véritable auteur du poème De balneis puteolanis,’ Société Nationale des Antiquaires de France 1 (1852) 334–53.Google Scholar

³ The editio princeps of his work was published by S. Engel, Petri d'Ebulo Carmen de motibus siculis (Basel 1746; repr. Naples 1770) and is based on the sole surviving MS (Bern Burgerbibliothek 120). The poem has been re-edited by E. Rocco with Italian translation (1845), Winkelmann, E. (1874), Rota, E. (1904), and Siragusa, G. B. (1906).Google Scholar

⁴ The Catalogus codicum manuscriptorum Bibliothecae Regis III/4 (Paris 1744) 434 pointed out that this manuscript was illustrated with figures: ‘Codex membranaceus, quo continetur Bolei, sive Ebolei, poëma de balneis, illorumque virtutibus: accedunt figurae minus elegantes. Is codex decimo quarto saeculo videtur exaratus.’ Note that in the index to the Catalogus the work is included as: ‘Anon. Poetae Carmina.’Google Scholar

⁵ The index to the Catalogue of Incipits conceals other MSS of the same poem. For instance, at col. 279 Thorndike and Kibre cite the incipit ‘Cum ad balnea veneris…’ for ‘De mirabilibus et balneis Puteolorum, printed at Naples, 1475, by Arnald of Brussels’: within this balneological treatise is, in fact, the editio princeps of portions of Peter's text. Cf. text to note 1 above. Similarly at col. 1393 under ‘Sciendum est quod a quampluribus scolaribus meis …’ is a reference to Ugolino da Monte Catini, De Balneis: Paris, Bibliothèque Nationale nouv. acq. lat. 211 (A.D. 1469) fols. 54^r–70^r. Ugolino in fact appended Peter's poem to his prose treatise completed in 1417 with the remark that he had obtained the poem from Bindacius de Ricasulis without being able to ascertain its provenance or the name of its author. Peter's poem is preserved by Ugolino in this MS on fols. 70^r--75^v (fols. 73^r--78^v according to a corrected numbering). The MS entered the Bibliothèque Nationale sometime between 1875 and 1891, as is indicated by the description of MSS for those dates in L. Delisle, Manuscrits latins et français … VII (Paris 1891) 229, quoting the incipit of this as-yet-unidentified Balnea puteolana as: ‘Inter opes etc.’ which seems not to have been noticed by Thorndike. It should be noted that Thorndike and Kibre are the first to notice, in their list of three MSS for the incipit ‘Inter opes,’ the fifteenth-century MS Oxford, Bodleian Library, Digby 129 fols. 29^v--(32^v) as a manuscript of Peter of Eboli's poem (col. 769).Google Scholar

⁶ De Marinis, T., ‘De balneis puteolanis,’ in: Classical, Mediaeval and Renaissance Studies in Honor of Berthold Louis Ullman (ed. C. Henderson, Jr.; Rome 1964) II 47–49 with 2 illus.Google Scholar

⁷ D'Amato, J. M., Prolegomena to a Critical Edition of the Illustrated Medieval Poem De balneis Terre Laboris by Peter of Eboli (diss. Johns Hopkins 1975) 273–74 (which refers to Peter's poem by an alternative title).Google Scholar

⁸ According to Thorndike, and Kibre, (col. 537), P belongs to the fifteenth century. But the Catalogue of the Royal Library (n. 4 above), Marinis, De (n. 6 above) 47, and D'Amato, (n. 7 above) 216 place it in the fourteenth, D'Amato assigning it to the third quarter. Imbault-Huart, M.-J. and Dubief, L., La médecine au moyen ǎge à travers les manuscrits de la Bibliothèque Nationale (Paris 1983) 170 assign it to the middle of the fourteenth century.Google Scholar

⁹ De Marinis, T., La biblioteca napoletana dei re D'Aragona (Milan 1947–52) I 145–61 and op. cit. (n. 6 above) 47. D'Amato's report (n. 7 above) 218 that the inscription appears on fol. i is a slip.Google Scholar

¹⁰ In collating P and m, I shall number the verses in accordance with the edition of 1553. But because this edition ends at verse 412 and some of the epigrams in P and m are thus missing, I have extended line-references to include: De balneo quod Arcus dicitur (413–24), De balneo quod Raynerii dicitur (425–36), and Balneum quod de Ferris dicitur (437–48). The formula for indicating titles varies considerably among the MSS. I have rendered the above three as they appear in P and m, and they are identical. For the sake of conciseness and the continuity of the argument, most of the variants which I shall mention have been relegated to the footnotes.Google Scholar

¹¹ Pm: 98 fugagat (for fugat); and reguma (for rheuma); 223 fuit vigilantiā om. (for fugit vigilantia lumina); 253 Crippa (for Cripta, the name of one of the baths); 281 horendaque (for horrendaque; one might expect this simple orthographic variant to appear in other MSS, but it does not); 290 febricitate (for febricitare); 291 dicti (for dici); 326d holeum (for oleum; again this orthographic variant does not appear in other MSS); 366 huc (for hunc); 370 illia (for ilia or ylia); 398 both MSS again have the name Crippa, but here in common with one other MS; 398 sautarem (for salutarem); 399 susape (for suscipe); 414 virtute (for virtutem); 416 aridia (for arida); 420 gecur (for iecur; but correctly at 112); 427 corporis (for corpus). To support her view that m is a copy of P, D'Amato (op. cit. n. 7 above) cites five readings, three of which are misreported: Pm do not read fugagit at 98, but fugagat; at 108 these MSS are not alone in reading nossunt for possunt, and m in fact reads not nossunt but nos sunt, which, if the right reading were not known from other MSS, might pass beyond suspicion; at 291 they read not ducti, but dicti. Besides these, she cites the readings at 399 and 437 included below.Google Scholar

¹² To give but a small sample: at 221 obtalpa and at 326c si morpheabit are read in Pm with two other MSS (Paris, Bibliothèque Nationale, franç. 1313, fols. v + 35 + v; and New York, Pierpont Morgan Library, Glazier 74, fols. ii + 35 + i). Other MSS at 221 read variously obtalmita, optarmita, ophtalmita, obtalmia, optalmia, opthalmia, octalmia, otalmia; and at 326c si morphea notabit (correctly), si morfea ledit, si morfea notatur, si morfea vocatur, sit formosa remota, sit formosa ve nota, and sit formosa venora). At 274 Nivem et, at 326d qua (for aqua) are read in Pm with one other MS. The rubricated titles Gregorius (for Georgius) and Petronio (for De Petroleo) are also errors in Pm with one other MS. Throughout this paper, when Pm are said to be in agreement with one other MS, it is always with the MS Morgan Library Glazier 74. At 274 other MSS read Ni vetet (correctly), Ni vetat, Ni vitet, Niviet, Ne vetet, Vivere et. At the heart of this error is a wrong word-division — m faithfully follows P in word-divisions even when wrong, as at 336 passioni, where Pm are uniquely in error for passio ni. Google Scholar

¹³ At 269, menstrua in P was miscopied into m as mensrua; at 289 ydropicos P became ydropigos m; at 296 mirificam written in full by P was miscopied as mirifica in m (since m copied from P exclusively, the fact that other MSS agree with m in having mirifica is no indication of their affinity to m); 299 frons P becomes fons m; 302 virga becomes virgo; 394 sumere becomes sumeret, the addition of t in m being made under the influence of the immediately following curet, as also in Morgan Library Glazier 74; 428 scabie becomes cabie; 430 utere becomes utero, doubtless under the influence of the immediately following name of the bath Raynerio; 435 refrenat becomes refrenant, as also in Morgan Library Glazier 74. I must emphasize that the errors peculiar to m here listed are also unique to m unless otherwise stated. There are also errors peculiar to P which m has inadvertently made worse. For example, at 275 the bath Silviana appears as silivana in P and silvana in m; the bath Braccula at 382 appears as bracula in P (here with other MSS) and bacula in m.Google Scholar

¹⁴ P had, for example, a penchant for representing final m by ӡ; hence at 394 Pm both have quisqӡ for quisquam. Google Scholar

¹⁵ For example at 218 and 264 in P is abbreviated as q'd in m (for quod).Google Scholar

¹⁶ For example at 167 from sӡ to set (showing a spontaneous preference of the scribe of m for set over sed).Google Scholar

¹⁷ Thus at 254 the abbreviation qӡ in P is expanded into que in m (an error for quia); at 290 P wrote prodē (for prodest), beside which a corrector (on whom see shortly below) marked a cross, and which was misunderstood by m as prodere; at 291 P wrote potit̄ instead of pot̄it, which m expanded into potitur — the scribe's error for poterit is explained by the misplaced stroke in its source P.Google Scholar

¹⁸ At 345, for example, he changes sed to set (cf. also n. 16 above); at 219 he changes vitium to vicium, at 273 gratia to gracia, at 310 sentiet to senciet, at 422 sotio to socio (here is another error unique to Pm, for scio); at 443 vitio to vicio, and frequently elsewhere t in P becomes c in m. Similarly at 100 decoquid P appears as dequoquid in m; at 191 sinthomata P becomes sinthomatha in m; sonpnus becomes sompnus at 223; rauca becomes raucha at 378.Google Scholar

¹⁹ At 263 consuesere in P is corrected to consuescere in m, and at 333 nervvos to nervos. At 326 olίί in m for in P looks like a conscious effort to correct, which does not arrive at the correct reading (olivi). Another instance (at 108) is mentioned below under a different classification.Google Scholar

²⁰ At 293 we find indie P, indice (in marg.) P², indice m; at 270 hec P, hoc (in marg.) P², hoc m; 312 [q]uid P, quod (in marg.) P², quod m; at 367 qua predicatis P, p̄latis (in marg.) P², quia prelatis m; at 414 mage P, magne (in marg.) P², magne m; at 426 salvi P, salsi (in marg.) P², salsi m; at 448 trino P, trina (in marg.) P², trina m. In all these cases m has the correct reading, but merely follows P² in each instance. Once, at 447 the scribe of P writes ephetat in the text and acknowledges a variant ‘aliter efferas’ in the margin, which the scribe of m reproduces exactly. Once too, at 437 we find vadis in the text of P, and both vatis and mitis by P² in the margin — m chooses mitis incorrectly. Peter is referring to a bath close not to any home on the shore of the ‘gentle’ (mitis) lake of Avernus, but to the home ‘of the Sibyl’ (vatis) on the shore of the lake. The mistaking of vatis for mitis suggests that the error arose in an exemplar written in Gothica rotunda.Google Scholar

²¹ We have already noticed (n. 17 above) that P² put a cross at 290, which failed to save m. At 99 P wrote edenti, above ti was written di by P², edendi m. Two other corrections attract special attention. At 260 cardinicam was written by P, ni was deleted in two separate strokes by P² thus leaving cardicam in error for cardiacam. Cardicam was written by m who followed P² in error. On the other hand, at 181 m reproduces quod P, not seeing over the d a faint cross superimposed by P². In this instance, m inadvertently fails to follow P², who had corrected quod to quo. In contrasting ways the unique errors in m at 260 and 181 are attributable to its exemplar and illustrate the dependence of m upon P.Google Scholar

²² In the first of these, a stroke through the first r in Palumbrara restores the correct name to this bath. The other precedes the bath Silviana and introduces a subdivision of the poem as follows: ‘Incipiunt balnea de avis (aquis P² supra lin.) salsis iuxta maris ripam existentia.’ The scribe of m, as usual, follows P² in both places. Avis for aquis is not an error peculiar to P, since it is also read by MS Morgan Library Glazier 74 to which P is most closely related. It is therefore highly improbable that P's avis is a lapsus calami which P² corrected from the exemplar. Nor is there any supporting evidence (as we shall confirm below) that P² made use of a second MS with which to check P. We can only therefore conclude that either the model of P also contained the correct reading as an alternative which P² added, or P² here emended, aquis being a simple enough emendation in the context.Google Scholar

²³ D'Amato, , op. cit. (n. 7 above) 217.Google Scholar

²⁴ Marinis, De, art. cit. (n. 6 above) 49.Google Scholar

²⁵ Just how faithfully is obvious at 255 (I indicate the initial letters of P² in square brackets), where the peculiar error of [u]unda (for [u]nda) in P is reproduced in m as Uunda, the initial U in gold. Among other unique errors in common of this kind, that of Uni in m, derived from a misreading of [a]ni in P, reveals in another way the direct dependence of m upon P², since a mistaking of a for u is common in Latin minuscule, but would not have occurred had m's exemplar had uncial letters as initials. Cf. at 130 [n]ec P, Nec m (for Hec); 422 [g]este P, Geste m (for Teste). Other errors of this category in Pm are shared with some other MSS. At 108 [n]ossunt (for Possunt) in P and MS Morgan Library Glazier 74 was ‘corrected’ in m to Nos sunt, the original error of an incorrect initial being now compounded by a wrong word-division. The following readings are also found in common with this single MS: at 429 [a]nfectam P, Anfectam m (for Infectam); 439 [s]ic P, Sic m (for Hic). It is significant for the following argument that at 328 Huius (for Cuius) and at 392 Sic (for Hic) are errors read in Pm together with three other MSS: Venice, Biblioteca Nazionale Marciana, Ital. Cl. XI. 124, fols. 54^v–62^v; and the two MSS mentioned in n. 12 above.Google Scholar

²⁶ Reproduced by Marinis, De, art. cit. (n. 6 above) pl. 1 facing p. 48, and Imbault-Huart, M.-J. and Dubief, L., op.cit. (n. 8 above) 171 pl. 77.Google Scholar

²⁷ Marinis, De, art. cit. (n. 6 above) 49. In fact, m was not written in several hands, but by a single scribe initially using black ink. There are a few corrections, but these seem to be by the scribe also. At 100, final -id was added later to dequoqu(id), in less luminously black ink than the surrounding text; similarly at 101 final -at was added to declar(at); and again at 219, final -i was added later to complete stomach(i) — in all three places space was left for the ending. At 157 quas was corrected to aquas by the insertion of a within the bowl of q; at 226 a stroke was added beneath the bowl of p to correct opatur into operatur; at 316, o is written above the first i of comiditatis — in all three instances in the same thin pen used for indicating initials in the margin. There is room for pause only at 119, where a final stroke turns longeve into longeva, the only change of reading from P.Google Scholar

²⁸ Noticed in connection with m by Marinis, De (n.9 above) II 254 doc. 351.Google Scholar

²⁹ Even odder, all the sections omitted by Marinis, De and two more (Raynerius and Trituli) are reported as missing by L. Petrucci, ‘Per una nuova edizione dei Bagni di Pozzuoli,’ Studi mediolatini e volgari 21 (1973) 238.Google Scholar

³⁰ It is worth mentioning, in passing, the existence of another MS, that now preserved in the Biblioteca Universitaria no. 860 at Valencia in Spain, which was produced in Naples in the same century in the Aragonese Scriptorium. Its scribe was Virgilio Ursuleo, who identifies himself on fol. 38^r. The manuscript must therefore have been written after 1455, when Ursuleo joined the royal Scriptorium, and before 1458, if De Marinis is correct in supposing that it was written for Ferdinand, who succeeded his father King Alfonso V upon his death in that year. Since it stayed in the library until the Franco-Spanish conquest of 1501, this manuscript was there at the same time as m was being copied from P. But it was not used as an exemplar from which to correct either P or m. As we have already concluded, P, P², and m derive from α-readings; the text of the Valencia MS belongs to another tradition. Cf. Marinis, De, art. cit (n. 6 above) 47 n. 3, where this MS is wrongly said to be no. 869, and also Kauffmann, op. cit. (n. 1 above) 82f. and De Marinis, op. cit (n. 9 above) 11.220 n. 601, referring to the inventory of MSS from Naples which eventually reached San Miguel de los Reyes in Spain, before passing to the University Library at Valencia.Google Scholar

³¹ Kauffmann, , op. cit. (n. 1 above) 38 remarks that ‘the Balneum Sancti Georgii composition is very different from its counterparts in the other manuscripts,’ but fails to recognize that it is actually the wrong miniature, being of the De Petra type. He instead attributes its incongruity to a different cause: ‘It is very probable that a late fifteenth-century artist of Cola Rapicano's stature should be prepared to take greater liberties in copying the composition of his model than were his predecessors.’ In fact, this illuminator of m stayed remarkably close to his model.Google Scholar

³² An asterisk in this table represents a bath omitted by Marinis, De in his list of contents. I have retained the spelling in Pm for the names of baths, correcting only that of St. George. The miniatures are not labelled in Pm, but have been identified by iconographical comparison with their counterparts in other illustrated MSS.Google Scholar